Feasibility of Estimating Type and Abundance of Iron Ore Using Field Hyperspectral Radiometry and Spectral Transformation in Precambrian Banded Iron Formations of North Odisha, India

In large-scale mining operations, methods need to be developed for rapid estimation of the relative abundance of iron without prior knowledge of the distribution of rock types. Imaging spectroscopy is a promising tool to identify the type of iron alteration mineral and its total abundance. In the present study, ore samples from nine Odisha mines were examined for spectral variability, feature characterization, and derivative analysis with the objective of finding the best predictor variable for total iron content in routine mine operation. Band depth maxima was found to be the best predictive variable by virtue of a strong correlation (R 2 = 898) with total iron content, the lowest RMSE (5.61), and the highest ME (0.899). Further, to identify the spectrally dominant mineral in the assemblage, spectral matching was performed using spectral angular mapper (SAM), spectral correlation angle (SCA), and spectral information divergence (SID). The overall accuracy of mineral identification is highest for SCA, but SAM was found to be the best matching algorithm for hematite.